Les;

OK, regarding cost concerns, I have thought this out a little but have not done thorough cost estimation.  A SqrOsc can be made with a 555 chip, four resistors and two caps plus two connectors.

Well, yes, if all you want is to go "beep". Oscillators in modular systems need a CV input for tracking; making sure that adding (typically) a extra Volt of modulation signal will double the frequency of the oscillator. This is hard to get stable for larger ranges. A CMOS chip like the 555 needs to be protected against getting inputs above it's power input or it will fry.

SqrOsc features like phase mod/output, hard-sync, FM modulation and a freq input are all available in hardware modules but those aren't cheap modules and none of those come close to the stability in tuning and tracking of this very modest UGen.

If you start making very cheap modules so people will buy large stacks of them to build long signal chains you'll need especially high quality modulation inputs because the longer the chain the more the errors will add up. Errors here will mean out of tune chords, for example. For oscillators you could use those cheap chips they used to use in telephone modems; those will do unusually stable FM over a huge frequency range; I'd go with those here but that doesn't get you off the hook for modulated filters.

My main reason for being sceptical about the practical feasibility of this is that people have tried it many times. Nearly all synthesis fanatics would like a system like the one you propose. Right now I think the cheapest series of modules on the market is made by the German company Doepfer; their modules will still set you back a few hundred bucks for the equivalent of a simple monosynth and I keep hearing about issues with their usage of relatively cheap connectors which will develop loose connections with use. Again; I love this idea as a dream but I fear that if it could be done it would've been done already.

I'd say the core of your plan is to represent signal flow as a 3d structure, right? I'd recommend going back to that and using a visual or physical representation of ChucK code in something like the Reactable or the Percussa Audiocubes.

To bring this back to the subject of ChucK; I'm not sure people realise that our young, incomplete, at times computationally ineficient and occasionally crash-prone language is immensenly stable and unfathomably powerfull compared to hardware modulars.

Yours,
Kas.